EMBRACE: A Multi-task Framework for Comprehensive Quality Assessment in Cleavage-stage Embryo

2026-07-10

AI Summary

This study addresses key challenges in cleavage-stage embryo assessment during in vitro fertilization—namely, inaccurate fragmentation detection, inconsistent developmental stage classification, and subjective symmetry scoring—by proposing the first multitask deep learning framework that integrates interpretable spatial fragment localization with holistic morphological analysis. Built upon a shared ResNet-50 backbone, the model incorporates a multiscale feature fusion module, a U-Net–style segmentation decoder, and task-specific classification heads to simultaneously perform cytoplasmic fragment segmentation, t2/t4 stage classification, and blastomere symmetry grading. On an independent test set, the framework achieves a Dice coefficient of 0.781 for segmentation, 0.995 accuracy for developmental stage classification, and 0.901 balanced accuracy for symmetry grading (weighted Kappa = 0.859), substantially enhancing the objectivity and consistency of embryo evaluation.

Abstract

Cleavage-stage embryo assessment in in vitro fertilization requires the integrated interpretation of cytoplasmic fragmentation, developmental stage, and blastomere symmetry. However, conventional visual assessment is affected by observer variability, particularly when fragmented regions are small, irregular, or low contrast. This study presents EMBRACE, a multi-task deep learning framework for jointly performing cytoplasmic-fragmentation segmentation, t2/t4 developmental-stage classification, and blastomere-symmetry grading from static cleavage-stage embryo microscopy images. EMBRACE combines a shared ResNet-50 backbone, a concatenation-based multi-scale feature-fusion (C-MSFF) module, a U-Net-style segmentation decoder, and two task-specific classification heads. After predefined inclusion and exclusion criteria, 9,137 annotated embryo images were divided into 7,309 training, 914 validation, and 914 held-out test images. On the held-out test set, EMBRACE achieved a Dice coefficient of 0.781 and an intersection over union of 0.677 for fragmentation segmentation. Developmental-stage classification achieved an accuracy of 0.995, macro-F1 of 0.994, and AUC of 1.000. Blastomere-symmetry grading achieved a balanced accuracy of 0.901, macro-F1 of 0.907, and quadratic weighted kappa of 0.859. These findings support the feasibility of combining spatially inspectable fragmentation localization with embryo-level morphology assessment in a single framework. External and prospective validation is required before clinical deployment.

Problem

• cleavage-stage embryo
• cytoplasmic fragmentation
• developmental stage
• blastomere symmetry
• embryo quality assessment

Innovation

• multi-task learning
• embryo quality assessment
• fragmentation segmentation
• multi-scale feature fusion
• deep learning